led driver and system reliability - department of energy–a product with mtbf of 50,000 hours will...
TRANSCRIPT
reliable l electrical l solutions
Warren Weeks, Director of Innovation & Technology Doug Hamilton, Director Research & Development Dennis Carpenter, Principal Reliability Engineer Chris Bailey, Director of Business Development
LED Driver and System Reliability February 3, 2016
reliable l electrical l solutions 2
All LED drivers are not created equal and need
to be used in their intended applications.
Premise
reliable l electrical l solutions
SSL Luminaire Burn-in
• 24 hour burn-in revealed limited driver mortality • AQL reduced burn-in time • 100% “Glow-n-Go” may be just as effective
stock photos Old school approach that is is expensive.
reliable l electrical l solutions
History of LED Driver Mortality (2011–2015)
• Blown fuses • Blown chokes • Failed MOVs • Damaged MOSFETs • Cracked solder joints • Damaged 0-10V dimming circuit
Not only electrolytic capacitors
reliable l electrical l solutions
Definition for LED driver / system failure? • Only if it generates a warranty claim? • Falls below DLC, Energy Star, CA Title 24 requirements? • Efficiency decrease? ROI is reduced? • Parametric performance drift?
• Increased ripple or flicker? • PF or THD out of tolerance?
• Mortality – no light output
reliable l electrical l solutions
Definition for Lifetime? • What is a failure? • What is the failure rate? • At what confidence level? • Ambient vs. case temperature? • Is max case temp important? • Is MTBF useful?
No industry consistency at all
reliable l electrical l solutions
• Original calculations defined by MIL-HDBK-217F
• Other similar commercial standards have evolved (Telcordia SR-332)
• Assumes constant failure rate for each component which may/is not be true
• For example:
– A product with MTBF of 50,000 hours will exhibit about 15% failures in the first year (8000 hours)
– A product with MTBF of 500,000 hours will exhibit about 1.5% failures in the first year (8000 hours)
7
MTBF = Mean Time Between Failure
Over-emphasis on MTBF as a measure of reliability is misleading and dangerous.
reliable l electrical l solutions 8
Expected Lifetime & Sources of Failures
“BATHTUB CURVE”
Component & Manufacturing Process Issues
“Random” Events that exceed component
specifications
True Wear-Out Mechanisms
reliable l electrical l solutions 9
Recommended Levels of Surge Protection
Usually requires added
surge protector device
120V only applications
120-277V light commercial Primarily 277V industrial
347/480V indoor
347/480V outdoor
{ {
Solid State Light Sources Used in Roadway and Area Lighting (ANSI C136.37-20xx)
reliable l electrical l solutions
Driver/System Reliability Methodology
• What causes system failures? – Elevated temperatures
– Thermal cycling
– Surge / transient events
– Repeat switching
• How to mimic this in “reasonable” amount of time to create failure distribution?
reliable l electrical l solutions
Driver Reliability Test Procedure
• 24 drivers and LED loads in environmental chamber
• Powered at maximum voltage: 277V, 347V, or 480V
• Cycled for 2 sec ON, 12 sec OFF
• 80C to -20C to 80C (repeat) – 3/min ramp time
– Relative Humidity ranges from 85% to 20%
– 1 hour dwell time at temperature
• Demonstrate 99% reliability at 90% confident level – Exponential distribution (constant failure rate) used to model zero (or low) failures
– Weibull analysis for failure distribution when sufficient failures occur
– 2 to 3 week test depending on number of failures
reliable l electrical l solutions
• 8 “Production” Luminaires
• Thermal Cycling (1 hour dwell time) – Outdoor fixtures: -20°C to 60°C
– Indoor fixtures: 15°C to 60°C
• Electrical Cycling – Fixtures powered as temp is raised,
de-energized as temper is lowered
– Minimum energized time of 1½ hours each cycle
• Humidity Cycling – 30% to 85% R.H., saturation period across multiple thermal cycles
– Promotes saturation and drying, accelerate potential aging of susceptible materials
• Single failure constitutes failure to meet requirements – Corrective action necessary to improve design or manufacturing process
12
Luminaire System Reliability Test Procedure
reliable l electrical l solutions
Parametric Performance
• Chroma C8000 automated system – 120V, 277V, 347V and 480V testing
• Full parametric pre-stress testing – Sets performance baseline
– Keep LED driver suppliers honest
• Repeat post-stress testing – Parametric out-of-spec “failure”
reliable l electrical l solutions
• LED driver selection should be very early in NPD process, not last
• Forget max case temperature; design to desired lifetime case temp
• Design surge protection for Location Category of worst environment
– Surge matters! And designs should be tested!
• Limit number of drivers per fixture – Failure rates are additive – FCC Title 47 CFR Part 15 troubles (remember to test) – How many LED drivers will be on sole dimming circuit? – Uses isolated electrical components
• Ask for driver manufacturers’ reliability test data and quality plan
14
LED Driver Selection – Rules of Thumb